How it works
Your app holds a knectar key. Your AI keys stay in one safe place.
Your app sends its knectar key. knectar checks it, adds your OpenAI key and sends the request on to OpenAI. If OpenAI fails, knectar can send it to a backup model of your choice.
what your app sends
Authorization: Bearer
Your app only ever holds the first one.
sk-bf-a91f sk-bf-7c02 sk-bf-e410
encrypted · last four shown
Your knectar key
Make one for each app. It goes where your OpenAI key used to go, like OPENAI_API_KEY in your .env. Keep it on your server, never in code that runs in the browser. The dashboard calls it a virtual key.
Your AI keys
When you create a knectar key, you paste in your OpenAI, Anthropic or other AI key. knectar stores it encrypted, tied to that one knectar key, and after that shows only its last four characters. Your app never sees it.
A backup model
Name a backup model in the request, or set one for each key in the dashboard. If the first AI provider fails, knectar sends the same request there. You write no retry code.
The usual setup puts your OpenAI key in every app. If one app leaks it, you make a new key and paste it into every app again.
With knectar, you switch off the knectar key that leaked, and nothing else breaks. Your AI keys never move.
One test request, sent through knectar by our server.
Our server sent this small request in the last ten minutes, using our own test key. These are the real results.
- gateway
- measuring
- model
- measuring
- answered in
- measuring
- result
- measuring
rflectr records what your coding agents send. knectar holds your AI keys.
rflectr runs on your computer and records every request your coding agents make. knectar runs in our cloud and keeps your AI keys out of your apps. Use them together or on their own.
See rflectr
